Part 21 Report - 1996-044

 
 

ACCESSION #: 9607170075



ANSYS [Registered        ANSYS,Inc.

  Trademark]             Johnson Road             Telephone 412.746.3304

                         Houston, PA 15342-1300   Facsimile 412.746.9494



                                   July 11, 1996



Thomas Greene, NRR/PECB

U .S. Nuclear Regulatory Commission

Mail Stop 0-11E4

Washington, DC 20555-0001



Ref: 1.  Accession 9601190322

     2.  ANSYS, Inc. letter dated June 24, 1996

     3.  Holtec Fax dated June 19, 1996

     4.  ANSYS, Inc. letter dated July 2, 1996



Subj:     Request for a Technical Review of a Draft Information Notice

          Regarding Erroneous Results from ANSYS 5.0A and ANSYS 5.1

          Computer Codes



Dear Mr.  Greene:



     After investigation of item 2 of Holtec's 10CFR21 Notification (i.e.

disparate results for forces and displacements are obtained depending

upon the techniques employed to launch the analysis runs), the following

was determined:



-    Using tho Holtec model as input, to ANSYS 5.1, minor changes in the

     results were noted for cases run on Personal Computers using

     different launch techniques.



-    No changes were found when using different launch techniques for

     cases run on workstations.



-    No changes were found when using different launch techniques to run

     ANSYS 5.2 on Personal Computers or workstations.



     Comparisons of printouts of the ANSYS database using both launch

methods yielded no unexpected differences to the G 15.8 format used.

Similarly, comparisons of printouts of the ANSYS model, including Fortran

COMMON blocks, using both launch methods yielded only differences that

were expected.



     Many changes were made to the ANSYS program between the 5.1 and 5.2

versions.  However, the changes in run time libraries may have

contributed the most in causing the reported differences in results. Root

cause analysis and extensive study of the outputs show no evidence of an

error in either version of



                ANSYS is a registered trademark of SASIP



Thomas Greene, NRR/PECB

U.S.  Nuclear Regulatory Commission

Page Two

July 11, 1996



ANSYS.  The reason for the reported differences obtained by using

different launch techniques is, again, the extreme numerical accuracy

required to solve Holtec's idealized problem.



     As was the case for item 1 of Holtec's notification, if perfect

symmetry is destroyed the problem is resolved and the launch method has

no effect on the results when using ANSYS Revision 5.1.



     The following summarizes our finding and recommendations on your

proposed draft notice:



     Our investigation of Holtec's model found a perfectly symmetric

     system that was inherently unstable, that physically could not

     exist, and that required extreme numerical precision to maintain the

     symmetry throughout the solution process.  A slight change in the

     model's dimensions resolves the problem immediately and results in

     repeatable solutions on all platforms, with all launch methods.



     We have found no evidence of an error in the ANSYS program and

therefore recommend that an Information Notice should not be published on

this matter.



                         Sincerely,



                         ANSYS, Inc.



                         William J. Bryan

                         Quality Assurance Manager



cc: Holtec, Inc.          ANSYS, Inc.

     A.  Soler              D.  Looman

                            D.  Conover

                            M.  Ingrund

                            J.  Bittner



*** END OF DOCUMENT ***